Research map / 3D vision / Face and shape
Learning Forgery-Aware Lip Representations Without Forgery Priors
Bofan Chen, Hongyu Zhu, Yi He, Sichu Liang, Shi-Lin Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace rotation pseudo-fakes with mixedfake lips (SML/CML) (ablation) | proposed detector · TFG-GRID PersonaTalk | AUC (%): 78.27 -> 99.92 | Improved |
| replace cross-entropy classifier with asymmetric contrastive loss (ablation) | proposed detector · TFG-GRID ER-NeRF | AUC (%): 62.83 -> 99.95 | Improved |
| replace SupCon with asymmetric contrastive loss (ablation) | proposed detector · TFG-GRID PersonaTalk | AUC (%): 65.90 -> 99.92 | Improved |
| replace SimCLR with asymmetric contrastive loss (ablation) | proposed detector · TFG-GRID ER-NeRF | AUC (%): 85.83 -> 99.95 | Improved |
| add Gaussian distribution matching loss (ablation) | proposed detector · AVLips (LOF, IF, OC-SVM, EE) | HTER (%): HTER reduced by 0.16 to 3.38 points | Improved |
| add human-impostor samples to training (V + HM) (ablation) | proposed detector · AVLips human-impostor attack | AUC (%): 95.24 -> 99.35 | Improved |
| add audio modality (AV variant) (ablation) | proposed detector · AVLips avg | AUC (%): 96.97 -> 99.84 | Improved |
| plug adapter into Siamese and SA-DTH encoders | Siamese, SA-DTH · TFG-GRID personalized TFGs | AUC (%): up to 40% AUC increase | Improved |
| train only on real data with no forgery priors vs supervised detectors | proposed detector · TFG-GRID avg | AUC (%): 95.26 -> 99.83 | Improved |
| default visual-only detector under real-world conditions, human-impostor attacks | proposed detector · AVLips human-impostor | AUC (%): TD-VSA 99.94 vs 95.24 | Mixed |
Closest papers by meaning
- SpeechForensics: Audio-Visual Speech Representation Learning for Face Forgery Detection (NeurIPS 2024 · similarity 0.90)
- Identity-Aware Vision-Language Model for Explainable Face Forgery Detection (AAAI 2026 · similarity 0.89)
- Lips Are Lying: Spotting the Temporal Inconsistency between Audio and Visual in Lip-Syncing DeepFakes (NeurIPS 2024 · similarity 0.88)
- Towards General Visual-Linguistic Face Forgery Detection (CVPR 2025 · similarity 0.87)
- Unmasking Puppeteers: Leveraging Biometric Leakage to Expose Impersonation in AI-Based Videoconferencing (NeurIPS 2025 · similarity 0.86)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
